arthur-pe / slicetca

Library to perform Slice Tensor Component Analysis (sliceTCA)

SliceTCA

This library provides tools to perform sliceTCA.

---

Installation

pip install slicetca

Full documentation

The full documentation can be found here.

Examples

Quick example

import slicetca
import torch
from matplotlib import pyplot as plt

device = ('cuda' if torch.cuda.is_available() else 'cpu')

# your_data is a numpy array of shape (trials, neurons, time).
data = torch.tensor(your_data, dtype=torch.float, device=device)

# The tensor is decomposed into 2 trial-, 0 neuron- and 3 time-slicing components.
components, model = slicetca.decompose(data, (2,0,3))

# For a not positive decomposition, we apply uniqueness constraints
model = slicetca.invariance(model)

slicetca.plot(model)

plt.show()

Notebook

See the example notebook for an application of sliceTCA to publicly available neural data.

Reference

A. Pellegrino_@^†, H. Stein^†, N. A. Cayco-Gaijc_@. (2024). Dimensionality reduction beyond neural subspaces with slice tensor component analysis. Nature Neuroscience https://www.nature.com/articles/s41593-024-01626-2.